rset/rsbetween.c

   1 /* $Id: rsbetween.c,v 1.12 2004-02-12 15:15:54 heikki Exp $
   2    Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002
   3    Index Data Aps
   4
   5 This file is part of the Zebra server.
   6
   7 Zebra is free software; you can redistribute it and/or modify it under
   8 the terms of the GNU General Public License as published by the Free
   9 Software Foundation; either version 2, or (at your option) any later
  10 version.
  11
  12 Zebra is distributed in the hope that it will be useful, but WITHOUT ANY
  13 WARRANTY; without even the implied warranty of MERCHANTABILITY or
  14 FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
  15 for more details.
  16
  17 You should have received a copy of the GNU General Public License
  18 along with Zebra; see the file LICENSE.zebra.  If not, write to the
  19 Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
  20 02111-1307, USA.
  21 */
  22
  23
  24 /* rsbetween is (mostly) used for xml searches. It returns the hits of the
  25  * "middle" rset, that are in between the "left" and "right" rsets. For
  26  * example "Shakespeare" in between "<title>" and </title>. The thing is
  27  * complicated by the inclusion of attributes (from their own rset). If attrs
  28  * specified, they must match the "left" rset (start tag). "Hamlet" between
  29  * "<title lang=eng>" and "</title>". (This assumes that the attributes are
  30  * indexed to the same seqno as the tags).
  31 */
  32
  33 #include <stdio.h>
  34 #include <stdlib.h>
  35 #include <string.h>
  36 #include <assert.h>
  37
  38 #include <rsbetween.h>
  39 #include <zebrautl.h>
  40
  41 static void *r_create_between(RSET ct, const struct rset_control *sel, void *parms);
  42 static RSFD r_open_between (RSET ct, int flag);
  43 static void r_close_between (RSFD rfd);
  44 static void r_delete_between (RSET ct);
  45 static void r_rewind_between (RSFD rfd);
  46 static int r_forward_between(RSET ct, RSFD rfd, void *buf, int *term_index,
  47                      int (*cmpfunc)(const void *p1, const void *p2),
  48                      const void *untilbuf);
  49 static int r_count_between (RSET ct);
  50 static int r_read_between (RSFD rfd, void *buf, int *term_index);
  51 static int r_write_between (RSFD rfd, const void *buf);
  52
  53 static const struct rset_control control_between =
  54 {
  55     "between",
  56     r_create_between,
  57     r_open_between,
  58     r_close_between,
  59     r_delete_between,
  60     r_rewind_between,
  61     r_forward_between, /* rset_default_forward, */
  62     r_count_between,
  63     r_read_between,
  64     r_write_between,
  65 };
  66
  67
  68 const struct rset_control *rset_kind_between = &control_between;
  69
  70 struct rset_between_info {
  71     int key_size;
  72     RSET rset_l;
  73     RSET rset_m;
  74     RSET rset_r;
  75     RSET rset_attr;
  76     int term_index_s;
  77     int (*cmp)(const void *p1, const void *p2);
  78     char *(*printer)(const void *p1, char *buf);
  79     struct rset_between_rfd *rfd_list;
  80 };
  81
  82 struct rset_between_rfd {
  83     RSFD rfd_l;
  84     RSFD rfd_m;
  85     RSFD rfd_r;
  86     RSFD rfd_attr;
  87     int  more_l;
  88     int  more_m;
  89     int  more_r;
  90     int  more_attr;
  91     int term_index_l;
  92     int term_index_m;
  93     int term_index_r;
  94     void *buf_l;
  95     void *buf_m;
  96     void *buf_r;
  97     void *buf_attr;
  98     int level;
  99     struct rset_between_rfd *next;
 100     struct rset_between_info *info;
 101 };
 102
 103 static void log2 (struct rset_between_rfd *p, char *msg, int cmp_l, int cmp_r)
 104 {
 105     char buf_l[32];
 106     char buf_m[32];
 107     char buf_r[32];
 108     logf(LOG_DEBUG,"btw: %s l=%s(%d/%d) m=%s(%d) r=%s(%d/%d), lev=%d",
 109       msg,
 110       (*p->info->printer)(p->buf_l, buf_l), p->more_l, cmp_l,
 111       (*p->info->printer)(p->buf_m, buf_m), p->more_m,
 112       (*p->info->printer)(p->buf_r, buf_r), p->more_r, cmp_r,
 113       p->level);
 114 }
 115
 116 static void *r_create_between (RSET ct, const struct rset_control *sel,
 117                                void *parms)
 118 {
 119     rset_between_parms *between_parms = (rset_between_parms *) parms;
 120     struct rset_between_info *info;
 121
 122     info = (struct rset_between_info *) xmalloc (sizeof(*info));
 123     info->key_size = between_parms->key_size;
 124     info->rset_l = between_parms->rset_l;
 125     info->rset_m = between_parms->rset_m;
 126     info->rset_r = between_parms->rset_r;
 127     info->rset_attr = between_parms->rset_attr;
 128     if (rset_is_volatile(info->rset_l) ||
 129         rset_is_volatile(info->rset_m) ||
 130         rset_is_volatile(info->rset_r))
 131         ct->flags |= RSET_FLAG_VOLATILE;
 132     info->cmp = between_parms->cmp;
 133     info->printer = between_parms->printer;
 134     info->rfd_list = NULL;
 135
 136     info->term_index_s = info->rset_l->no_rset_terms;
 137     if (info->rset_m)
 138     {
 139         ct->no_rset_terms =
 140             info->rset_l->no_rset_terms +
 141             info->rset_m->no_rset_terms +
 142             info->rset_r->no_rset_terms;
 143         ct->rset_terms = (RSET_TERM *)
 144             xmalloc (sizeof (*ct->rset_terms) * ct->no_rset_terms);
 145         memcpy (ct->rset_terms, info->rset_l->rset_terms,
 146                 info->rset_l->no_rset_terms * sizeof(*ct->rset_terms));
 147         memcpy (ct->rset_terms + info->rset_l->no_rset_terms,
 148                 info->rset_m->rset_terms,
 149                 info->rset_m->no_rset_terms * sizeof(*ct->rset_terms));
 150         memcpy (ct->rset_terms + info->rset_l->no_rset_terms +
 151                 info->rset_m->no_rset_terms,
 152                 info->rset_r->rset_terms,
 153                 info->rset_r->no_rset_terms * sizeof(*ct->rset_terms));
 154     }
 155     else
 156     {
 157         ct->no_rset_terms =
 158             info->rset_l->no_rset_terms +
 159             info->rset_r->no_rset_terms;
 160         ct->rset_terms = (RSET_TERM *)
 161             xmalloc (sizeof (*ct->rset_terms) * ct->no_rset_terms);
 162         memcpy (ct->rset_terms, info->rset_l->rset_terms,
 163                 info->rset_l->no_rset_terms * sizeof(*ct->rset_terms));
 164         memcpy (ct->rset_terms + info->rset_l->no_rset_terms,
 165                 info->rset_r->rset_terms,
 166                 info->rset_r->no_rset_terms * sizeof(*ct->rset_terms));
 167     }
 168
 169     return info;
 170 }
 171
 172 static RSFD r_open_between (RSET ct, int flag)
 173 {
 174     struct rset_between_info *info = (struct rset_between_info *) ct->buf;
 175     struct rset_between_rfd *rfd;
 176
 177     if (flag & RSETF_WRITE)
 178     {
 179         logf (LOG_FATAL, "between set type is read-only");
 180         return NULL;
 181     }
 182     rfd = (struct rset_between_rfd *) xmalloc (sizeof(*rfd));
 183     rfd->next = info->rfd_list;
 184     info->rfd_list = rfd;
 185     rfd->info = info;
 186
 187     rfd->buf_l = xmalloc (info->key_size);
 188     rfd->buf_m = xmalloc (info->key_size);
 189     rfd->buf_r = xmalloc (info->key_size);
 190     rfd->buf_attr = xmalloc (info->key_size);
 191
 192     rfd->rfd_l = rset_open (info->rset_l, RSETF_READ);
 193     rfd->rfd_m = rset_open (info->rset_m, RSETF_READ);
 194     rfd->rfd_r = rset_open (info->rset_r, RSETF_READ);
 195
 196     rfd->more_l = rset_read (info->rset_l, rfd->rfd_l, rfd->buf_l,
 197                              &rfd->term_index_l);
 198     rfd->more_m = rset_read (info->rset_m, rfd->rfd_m, rfd->buf_m,
 199                              &rfd->term_index_m);
 200     rfd->more_r = rset_read (info->rset_r, rfd->rfd_r, rfd->buf_r,
 201                              &rfd->term_index_r);
 202     if (info->rset_attr)
 203     {
 204         int dummy;
 205         rfd->rfd_attr = rset_open (info->rset_attr, RSETF_READ);
 206         rfd->more_attr = rset_read (info->rset_attr, rfd->rfd_attr,
 207                                     rfd->buf_attr, &dummy);
 208     }
 209     rfd->level=0;
 210     return rfd;
 211 }
 212
 213 static void r_close_between (RSFD rfd)
 214 {
 215     struct rset_between_info *info = ((struct rset_between_rfd*)rfd)->info;
 216     struct rset_between_rfd **rfdp;
 217
 218     for (rfdp = &info->rfd_list; *rfdp; rfdp = &(*rfdp)->next)
 219         if (*rfdp == rfd)
 220         {
 221             xfree ((*rfdp)->buf_l);
 222             xfree ((*rfdp)->buf_m);
 223             xfree ((*rfdp)->buf_r);
 224             xfree ((*rfdp)->buf_attr);
 225             rset_close (info->rset_l, (*rfdp)->rfd_l);
 226             rset_close (info->rset_m, (*rfdp)->rfd_m);
 227             rset_close (info->rset_r, (*rfdp)->rfd_r);
 228             if (info->rset_attr)
 229                 rset_close (info->rset_attr, (*rfdp)->rfd_attr);
 230
 231             *rfdp = (*rfdp)->next;
 232             xfree (rfd);
 233             return;
 234         }
 235     logf (LOG_FATAL, "r_close_between but no rfd match!");
 236     assert (0);
 237 }
 238
 239 static void r_delete_between (RSET ct)
 240 {
 241     struct rset_between_info *info = (struct rset_between_info *) ct->buf;
 242
 243     assert (info->rfd_list == NULL);
 244     xfree (ct->rset_terms);
 245     rset_delete (info->rset_l);
 246     rset_delete (info->rset_m);
 247     rset_delete (info->rset_r);
 248     if (info->rset_attr)
 249         rset_delete (info->rset_attr);
 250     xfree (info);
 251 }
 252
 253 static void r_rewind_between (RSFD rfd)
 254 {
 255     struct rset_between_info *info = ((struct rset_between_rfd*)rfd)->info;
 256     struct rset_between_rfd *p = (struct rset_between_rfd *) rfd;
 257
 258     logf (LOG_DEBUG, "rsbetween_rewind");
 259     rset_rewind (info->rset_l, p->rfd_l);
 260     rset_rewind (info->rset_m, p->rfd_m);
 261     rset_rewind (info->rset_r, p->rfd_r);
 262     p->more_l = rset_read (info->rset_l, p->rfd_l, p->buf_l, &p->term_index_l);
 263     p->more_m = rset_read (info->rset_m, p->rfd_m, p->buf_m, &p->term_index_m);
 264     p->more_r = rset_read (info->rset_r, p->rfd_r, p->buf_r, &p->term_index_r);
 265     if (info->rset_attr)
 266     {
 267         int dummy;
 268         rset_rewind (info->rset_attr, p->rfd_attr);
 269         p->more_attr = rset_read (info->rset_attr, p->rfd_attr, p->buf_attr,
 270                                   &dummy);
 271     }
 272     p->level=0;
 273 }
 274
 275
 276
 277 static int r_forward_between(RSET ct, RSFD rfd, void *buf, int *term_index,
 278                      int (*cmpfunc)(const void *p1, const void *p2),
 279                      const void *untilbuf)
 280 {
 281     struct rset_between_info *info = ((struct rset_between_rfd*)rfd)->info;
 282     struct rset_between_rfd *p = (struct rset_between_rfd *) rfd;
 283     int rc;
 284     log2( p, "fwd: before forward", 0,0);
 285     /* It is enough to forward the m pointer here, the read will */
 286     /* naturally forward the l, m, and attr pointers */
 287     if (p->more_m)
 288         p->more_m=rset_forward(info->rset_m,p->rfd_m, p->buf_m,
 289                         &p->term_index_m, info->cmp,untilbuf);
 290     log2( p, "fwd: after forward M", 0,0);
 291     rc = r_read_between(rfd, buf, term_index);
 292     log2( p, "fwd: after forward", 0,0);
 293     return rc;
 294 }
 295
 296 static int r_count_between (RSET ct)
 297 {
 298     return 0;
 299 }
 300
 301
 302
 303 static int r_read_between (RSFD rfd, void *buf, int *term_index)
 304 {
 305     struct rset_between_rfd *p = (struct rset_between_rfd *) rfd;
 306     struct rset_between_info *info = p->info;
 307     int cmp_l=0;
 308     int cmp_r=0;
 309     int attr_match = 0;
 310
 311     while (p->more_m)
 312     {
 313         log2( p, "start of loop", cmp_l, cmp_r);
 314
 315         /* forward L until past m, count levels, note rec boundaries */
 316         if (p->more_l)
 317             cmp_l= (*info->cmp)(p->buf_l, p->buf_m);
 318         else
 319         {
 320             p->level = 0;
 321             cmp_l=2; /* past this record */
 322         }
 323         log2( p, "after first L", cmp_l, cmp_r);
 324
 325         while (cmp_l < 0)   /* l before m */
 326         {
 327             if (cmp_l == -2)
 328                 p->level=0; /* earlier record */
 329             if (cmp_l == -1)
 330             {
 331                 p->level++; /* relevant start tag */
 332
 333                 if (!info->rset_attr)
 334                     attr_match = 1;
 335                 else
 336                 {
 337                     int cmp_attr;
 338                     int dummy_term;
 339                     attr_match = 0;
 340                     while (p->more_attr)
 341                     {
 342                         cmp_attr = (*info->cmp)(p->buf_attr, p->buf_l);
 343                         if (cmp_attr == 0)
 344                         {
 345                             attr_match = 1;
 346                             break;
 347                         }
 348                         else if (cmp_attr > 0)
 349                             break;
 350                         else if (cmp_attr==-1)
 351                             p->more_attr = rset_read (info->rset_attr, p->rfd_attr,
 352                                                   p->buf_attr, &dummy_term);
 353                             /* if we had a forward that went all the way to
 354                              * the seqno, we could use that. But fwd only goes
 355                              * to the sysno */
 356                         else if (cmp_attr==-2)
 357                         {
 358                             p->more_attr = rset_forward(
 359                                       info->rset_attr, p->rfd_attr,
 360                                       p->buf_attr, &dummy_term,
 361                                       info->cmp, p->buf_l);
 362                             logf(LOG_DEBUG, "btw: after frowarding attr m=%d",p->more_attr);
 363                         }
 364                     } /* while more_attr */
 365                 }
 366             }
 367 #define NEWCODE 1
 368 #if NEWCODE
 369             if (cmp_l==-2)
 370             {
 371                 if (p->more_l)
 372                 {
 373                     p->more_l=rset_forward(
 374                                       info->rset_l, p->rfd_l,
 375                                       p->buf_l, &p->term_index_l,
 376                                       info->cmp, p->buf_m);
 377                     if (p->more_l)
 378                         cmp_l= (*info->cmp)(p->buf_l, p->buf_m);
 379                     else
 380                         cmp_l=2;
 381                     log2( p, "after forwarding L", cmp_l, cmp_r);
 382                 }
 383             } else
 384             {
 385                 p->more_l = rset_read (info->rset_l, p->rfd_l, p->buf_l,
 386                               &p->term_index_l);
 387             }
 388 #else
 389             p->more_l = rset_read (info->rset_l, p->rfd_l, p->buf_l,
 390                               &p->term_index_l);
 391 #endif
 392             if (p->more_l)
 393             {
 394                 cmp_l= (*info->cmp)(p->buf_l, p->buf_m);
 395             }
 396             else
 397                 cmp_l=2;
 398             log2( p, "end of L loop", cmp_l, cmp_r);
 399         } /* forward L */
 400
 401
 402         /* forward R until past m, count levels */
 403         log2( p, "Before moving R", cmp_l, cmp_r);
 404         if (p->more_r)
 405             cmp_r= (*info->cmp)(p->buf_r, p->buf_m);
 406         else
 407             cmp_r=2;
 408         log2( p, "after first R", cmp_l, cmp_r);
 409         while (cmp_r < 0)   /* r before m */
 410         {
 411              /* -2, earlier record, don't count level */
 412             if (cmp_r == -1)
 413                 p->level--; /* relevant end tag */
 414             if (p->more_r)
 415             {
 416 #if NEWCODE
 417                 if (cmp_r==-2)
 418                 {
 419                     p->more_r=rset_forward(
 420                                       info->rset_r, p->rfd_r,
 421                                       p->buf_r, &p->term_index_r,
 422                                       info->cmp, p->buf_m);
 423                 } else
 424                 {
 425                     p->more_r = rset_read (info->rset_r, p->rfd_r, p->buf_r,
 426                                        &p->term_index_r);
 427                 }
 428                 if (p->more_r)
 429                     cmp_r= (*info->cmp)(p->buf_r, p->buf_m);
 430
 431 #else
 432                 p->more_r = rset_read (info->rset_r, p->rfd_r, p->buf_r,
 433                                        &p->term_index_r);
 434                 cmp_r= (*info->cmp)(p->buf_r, p->buf_m);
 435 #endif
 436             }
 437             else
 438                 cmp_r=2;
 439         log2( p, "End of R loop", cmp_l, cmp_r);
 440         } /* forward R */
 441
 442         if ( ( p->level <= 0 ) && ! p->more_l)
 443             return 0; /* no more start tags, nothing more to find */
 444
 445         if ( attr_match && p->level > 0)  /* within a tag pair (or deeper) */
 446         {
 447             memcpy (buf, p->buf_m, info->key_size);
 448             *term_index = p->term_index_m;
 449             log2( p, "Returning a hit (and forwarding m)", cmp_l, cmp_r);
 450             p->more_m = rset_read (info->rset_m, p->rfd_m, p->buf_m,
 451                                    &p->term_index_m);
 452             if (cmp_l == 2)
 453                 p->level = 0;
 454             return 1;
 455         }
 456         else if ( ! p->more_l )  /* not in data, no more starts */
 457         {
 458             log2( p, "no more starts, exiting without a hit", cmp_l, cmp_r);
 459             return 0;  /* ergo, nothing can be found. stop scanning */
 460         }
 461 #if NEWCODE
 462         if (cmp_l == 2)
 463         {
 464             p->level = 0;
 465             p->more_m=rset_forward(
 466                               info->rset_m, p->rfd_m,
 467                               p->buf_m, &p->term_index_m,
 468                               info->cmp, p->buf_l);
 469         } else
 470         {
 471             p->more_m = rset_read (info->rset_m, p->rfd_m, p->buf_m,
 472                                &p->term_index_m);
 473         }
 474 #else
 475         if (cmp_l == 2)
 476             p->level = 0;
 477         p->more_m = rset_read (info->rset_m, p->rfd_m, p->buf_m,
 478                                &p->term_index_m);
 479 #endif
 480         log2( p, "End of M loop", cmp_l, cmp_r);
 481     } /* while more_m */
 482
 483     log2( p, "Exiting, nothing more in m", cmp_l, cmp_r);
 484     return 0;  /* no more data possible */
 485
 486
 487 }  /* r_read */
 488
 489
 490 static int r_write_between (RSFD rfd, const void *buf)
 491 {
 492     logf (LOG_FATAL, "between set type is read-only");
 493     return -1;
 494 }
 495